By Assi Hansen, Isamu Miyamoto, Tiina Amberla, Yasuhiro Okamoto

One of the most employed material combinations in microtechnology is the glass-silicon pair. Since the huge growth of the microelectromechanical devices, sensors and micro-fluidic devices, bonding of this couple has become more and more critical issue. Many of these glass-silicon bonding processes, for example anodic bonding, fusion bonding and eutectic bonding, have reached their limits in terms of flexibility. In addition, high temperature and electrical field applied in the process can limit the application. Compared to conventional methods, local laser joining technique can provide many advantages, such as localized and flexible joining, non-contact manufacturing, high precision and repeatability, easiness and minimization of heat effects. The technique is based on the principle of transmission welding, where glass is transparent to the wavelength used and hence the laser beam passes through the glass wafer and is absorbed to silicon. As a result, silicon is melted and upon resolidification bonding is realized between the two substrates.

We present study on the local laser joining of two different glass types to silicon. Glass materials were chosen to be Borofloat33 due to the commercial use and SW-Y due to the best thermal properties. Laser bonding tests were made using picosecond pulsed fiber laser (X-LASE, Corelase, Finland) with central wavelength of 1070 nm and pulse duration of 20 ps. Focusing optics of 30 lens was used. Samples were placed into X-Y -table either without any pressuring system or fixture vacuum jig.

Optical contact area between the silicon and glass was reduced close to weld bead width since the large optical contact area was influencing for the shearing load. Etched silicon plates and flat glass plates were carefully cleaned and then brought into intimate contact before the actual laser bonding process. Focused laser beam was irradiated to the interface from the glass wafer side. Numerous parameter tests were done during the study using different repetition rates (0.25-4 MHz), velocities (500-2000 mm/s), and powers (0.125-16 W). Based on the different parameter tests, some of the parameters were chosen for the other tests e.g., scanning electron microscopic and elementary analysis, mechanical tests and accelerated life tests.

Through the experiments, it was found out that these two different glass types and silicon wafers could be local joined successfully with high quality, small heat input, and hermetically. For all tests, it was crucial that the laser beam was focused properly into the interface of glass and silicon and that the optical contact area between the silicon and glass was formed well before the actual laser joining.

During the laser joining process, the anchor shape was noticed with some parameters and Si-glass combinations. Under high velocity condition, some particles of silicon remained which indicated of the convection. Under low velocity condition, particles of silicon were not remarkable which indicated of the diffusion. No-gap and 3D structure in mixture area might lead tight joint. For testing the temperature and humidity, the enclosed samples were needed to be sure of the hermetic behavior. The test was done according to JEDEC’s JESD22-A104C “Temperature Cycling” standard. The minimum and maximum temperature limits were -40 ˚C and +85 ˚C. Samples remained in good condition during the accelerated life test. No damage was noticed in samples.

Many shearing tests were carried out for different pulse overlap rate and scanning velocity condition. Shear loads were almost the same as joining the samples with (~46 MPa) or without the vacuum jig (~47 MPa). Vacuum jig was not absolutely necessary conditions. In the case of Borofloat33, the tendency of shearing stress for number of pulse was the same as SW-Y, while the strength of Borofloat33 (~45 MPa) was weaker than SW-Y (~60 MPa). A suitable pulse repetition rate depends on the pulse overlap rate, and a suitable pulse overlap rate should be used at the certain pulse repetition rate in order to obtain the higher shearing strength. On the other hand, if the peak of shearing stress would shift the higher pulse number, the high pulse repetition rate becomes effective.

The following images include Silicon-glass(Borofloat33) laser joined sample and diamond saw cut chips, SEM-profile picture and cross section of the interface, respectively.

1st figure

2nd figure

Si-glass chip  Si-glass chip

4th figure